Different CMFs with Spectros

Hi Steve,

Thanks for clarifying. I was wondering if it might be due to the restriction to using only the XYZ data that X-Rite officially supports, rather than the full spectral data that's accessible via some open-source tools like Argyll. Thanks

Spectral data is not really of use for calibration - it needs to be converted into XYZ/Yxy data.
So for probes that have that as their standard output, it makes sense to use that, and no bother with spectral data.

However, we are going to add Spectral Graphs to ColourSpace, so at that point we will natively use spectral data too.

Steve

Steve

That's great news about ColourSpace natively using spectral data!

Perhaps the original intention of this thread was to say that since XYZ/Yxy values are caculated by multiplying the measured spectral power distribution by the CMF function and then integrating. Having direct access to the spectral power distribution would allow for the flexibility of applying different CMFs (such as the CIE 1964 10° CMF) to convert into XYZ/Yxy values accordingly .

There will be no difference between calculating XYZ by colourspace and calculating XYZ by the spectrometer internal program. My guess is that calculating XYZ will slow down the operation of ColourSpace or increase instability.
Not all spectrometers provide CMF switching function, and software calculation can make up for this.
As a leading color science software, I think CS should not be limited to XYZ but trace back to the source of XYZ and calculate everything.

We've had a couple of questions on CMFs recently, probably due to Calman stating they now include CIE 2015 CMFs...

The issue is that CMFs can only be used with Spectral data when converting into XYZ data, as used by most calibration systems.
This has been discussed within the thread already, but it is worth reiterating that trying to apply CMFs to existing XYZ measurement data is not a valid workflow, and will generate inaccurate results.
When using Spectros, many have the ability to internally apply a given CMS to the measured spectral data BEFORE exporting standard XYZ data to the calibration program.
This can be seen with CR-250/300 Spectros for example.

With Colourimeters, the action of Probe Matching effectively burns into the probe measurements the CMF that was in operation when the Spectro was use to generate the Probe Matching file.
With ColourSpace this can uniquely be applied retrospectively, rather than being burnt-in to the Colourimeter measurements.

As and when we add Spectral Data capabilities to ColourSpace we will at that point look at adding User Selective CMFs, so they operate correctly.
We will not add fake CMF options to recorded XYZ data.

Steve

Any plans to add Spectral Data capabilities to ColourSpace in the near future?

We have lots of plans, but don't comment on future developments are that not likely to be in the next release.

Steve

For individuals, visual matching is far superior to any universal CMF method (universal CMF includes CIE 1931, CIE 170, Apple CMF, etc.). For industrial production, universal CMF is the only option.
Until methods and instruments for measuring individual CMF become available, I do not trust any universal CMF. They are sometimes effective, and sometimes worse than 1931.

Then why are you querying Admesy Hera for spectral data on every measurement?

The code is in there for all (most?) spectros, ready for when we do.

As stated, presently we do not use spectral data.
The only before would be for spectral graphs.

Steve

Hi
You're absolutely right!

The individual CMFs for display colour management is becoming increasingly important, particularly given the trend towards ever-narrower-band primary colours, such as those found in laser projectors.

ICC held an Expert Day on Individual Colour Matching Functions, in conjunction with the CIE Scientific Conference in Vienna on July 8 2025.
https://www.color.org/events/vienna/

The goal of the meeting was to review recent work on individual colour matching functions and its application to colour management.

Andrew Stockman gave a presentation at the ICC meeting in which he provided a clear overview of the current state of individual CMF:
Cone spectral sensitivities, colour matching functions and individual differences Andrew Stockman, University College London
https://www.color.org/events/vienna/1.AndrewStockman_ICC_LEDMax_ICMFs.pdf

Stockman has developed a template to correct for individual observer metamerism:
"Formulae for generating standard and individual human cone spectral sensitivities"
https://onlinelibrary.wiley.com/doi/full/10.1002/col.22879

As the standard observer defined in CIE 170 2015_2 is an adaptation to the entire population,
"To produce the mean L-cone spectral sensitivity subsequently adopted by the CIE, Stockman
and Sharpe linearly combined the L(ala180) and L(ser180) spectral sensitivities in the ratio of 0.44:0.56 (based on the proportion of polymorphisms found in the population)",
it is never matching for each individual, independent on which type of observer they belong to.

Since I was never entirely satisfied with the white point, even after calibrating my Eizo 319 using Probe match with the spectrometer, I used the template to create my individual CMF!

With the script https://github.com/CVRL-IoO/Individual-CMFs you can modify the standard 2015 observer.

Stockman has incorporated further factors, such as the optical density of the lens—which is age-dependent—as possible parameters into his template for the individual CMF.

As I'm 61 years old I have addressed this in the script to compensate the higher optical density of my lens according to CIE 16x:2005:
"Table 6.10 The optical density Docul(λ) of lens and other ocular media as a function of age"
(Stockman, Sharpe and Fach (1999).

To get an idea about my colour vision and my individual shifts L(ala180) or L(ser180) for the implementing into my individual CMF I have tested this with the instrument Anomaloskop: oculus https://www.oculus.de/en/products/hmc-anomaloskop/?highlights

With the native linear quanta Data from the template (without normalisation) and the provided 3x3 conversation Matrix LMS to XYZ of the paper https://onlinelibrary.wiley.com/doi/full/10.1002/col.22879

I got my individual CMF!

I used this individual CMF with my spectrometer and set the monitor's white point using the Interactive Controls of my Eizo to D65. Then I used colourspace to create a new Colour Space Target with this white point.

Then I create a 3D calibration LUT as normal with ColourSpace (re-setting the white point in colourspace to D65 for the profile) , selecting the new Colour Space Target as the Source.
https://lightillusion.com/perceptual_match_guide.html

This procedure has generated a custom spectral matching to my individual colour vision.

For the first time I have not the feeling that my display has yellow tint because my individual CMF has compensated my higher lens density and my higher sensitivity to the longer wavelength compared to the standard 2015 observer.

Best

Frank